Re^2: Overloading print()

... could you perhaps share the situation you're in where it would be handy for these two to behave differently?

Firstly, I see what you mean by the expectation that print $blessed_obj; and print "$blessed_obj"; should do the same thing. (A nicely insightful observation :-)

My question came about this way:
In C, users of the mpfr library will output values to stdout using the mpfr_out_str function - which prints directly to stdout (rather than return a string which one can then print to stdout). As regards Math::MPFR (which wraps the mpfr library function), I thought "Wouldn't it be nice if print was overloaded to use mpfr_out_str ... then C and Perl are guaranteed of producing the same output." But then, that would mean that every time someone did $x = "$blessed_obj"; the value encapsulated in $blessed_obj is going to be printed to stdout ... and that's probably not what's intended :-)

Anyway, as you've just demonstrated, there's a very good reason that print and "" should be overloaded by the same function ... so I think that's the end of that train of thought. Probably better to have print and "" overloaded by a function that returns a string that emulates the mpfr_out_str output - as is currently the case. Only problem there is that, since the mpfr library itself doesn't provide such a function, I have to hand craft my own ... and therein lies the potential for discrepancies.

Thanks jbert.

Cheers,
Rob

Comment on Re^2: Overloading print() Select or Download Code

Replies are listed 'Best First'.
Re^3: Overloading print() by moritz (Cardinal) on Sep 05, 2007 at 09:46 UTC
Why don't you make Math::MPFR capture the output to STDOUT by localizing *STDOUT in the wrapper functions and redirecting it to a string? Then you can return the strings instead of printing them, and produce all output with perl's print. Or do you do that already?	[reply]
Re^4: Overloading print() by syphilis (Archbishop) on Sep 05, 2007 at 09:51 UTC
Or do you do that already? Nope ... but it sounds promising. Where do I find the relevant documentation ? (Haven't done anything like this before, afaicr.) Cheers, Rob	[reply]
Re^5: Overloading print() by moritz (Cardinal) on Sep 05, 2007 at 10:52 UTC
it's basically in open and local, an example is in the answers to Redirecting STDOUT, lexically scoped Perl 6 in German -- Difficult Sudoku	[reply]
Re^6: Overloading print() by syphilis (Archbishop) on Sep 05, 2007 at 12:36 UTC
Re^7: Overloading print() by BrowserUk (Patriarch) on Sep 05, 2007 at 12:54 UTC
Re^3: Overloading print() by jbert (Priest) on Sep 05, 2007 at 10:15 UTC
Cool, thanks for the explanation. So...it sounds as though the problem is problematic API. So the real fix is to refactor the underlying MPFR "output string" func to call a 'format to string' helper function, which you can then use for your stringification. I appreciate that might not be easy/desirable/practical (mimicing the func in perl is probably easier), but hey.	[reply]
Re^4: Overloading print() by syphilis (Archbishop) on Sep 05, 2007 at 12:57 UTC
... the real* fix is to refactor the underlying MPFR "output string" func to call a 'format to string' helper function* At which point syphilis is thankful that it's impossible to panic oneself to death .... (Anyway, you're wrong ... the real fix is to delegate responsibility :-))))))) Cheers, Rob	[reply]
Re^3: Overloading print() by pilcrow (Sexton) on Sep 05, 2007 at 17:28 UTC
"Wouldn't it be nice if print was overloaded to use mpfr_out_str" ... But then, that would mean that every time someone did $x = "$blessed_obj"; the value encapsulated in $blessed_obj is going to be printed to stdout `mpfr_get_str()` probably does what you want here. If so, I'd wager that Math::MPFR already does what you want, since it exposes `Rmpfr_get_str()` and, using it, already overloads its objects' stringification. -pilcrow	[reply] [d/l] [select]
Re^4: Overloading print() by syphilis (Archbishop) on Sep 06, 2007 at 05:52 UTC
mpfr_get_str() probably does what you want here Unfortunately it doesn't. All it does is return a string of digits (with a leading minus sign if it's a negative number) for the mantissa, and sets a variable to the value of the exponent. It's then up to `Rmpfr_get_str` to put the number into scientific notation, based on the information provided by `mpfr_get_str`. If, as one would instinctively expect, `mpfr_get_str` returned a string formatted in the same way as the `mpfr_out_str` output, then you would be absolutely correct ... and this whole thread would never bave been born :-) Cheers, Rob	[reply] [d/l] [select]